Questions about Joy
Mark Phillips — 2003-11-19 04:17:25
Hi,
I've just heard about Joy and it sounds like a
very interesting language. I have a few questions
which I'm hoping people can shed some light on.
1. Is Joy an exploratory language only? Could it
be used in a real-world application? If not, how
far away from this is it? Could it be used in
combination with another language to make it
useable?
2. Is there potential eventually for a sophisticated
type system as part of Joy? Would any real typing
system have to be modelled within the Joy language
itself? Is this a reasonable thing to do?
3. How are sophisticated data structures best
modelled in Joy?
4. Do people see concatenative languages as
THE way forward, or do they see them as a good
tool to use for some types of programming?
5. Do notions of parallel programing fit within
Joy's model?
6. What, if any, are the downsides of Joy's
approach?
Thanks,
Mark.
__________________________________
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
http://antispam.yahoo.com/whatsnewfree
Slava Pestov — 2003-11-19 04:54:40
On Tue, 2003-11-18 at 23:17, Mark Phillips wrote:
> 1. Is Joy an exploratory language only? Could it
> be used in a real-world application? If not, how
> far away from this is it? Could it be used in
> combination with another language to make it
> useable?
Well, right now, I'm writing an action game in Java; for scripting, it
uses a language I'm developing that's closely modelled after Joy. Its
not exactly the same as Joy, though:
- more Forth-like stack operators: over, nip, tuck, rot, -rot, 2drop,
2dup, and so on; although there is a 'dip' like in Joy as well.
- variables (hence side effects) -- $foo pushes value of foo, @foo sets
value of foo to the top of the stack (actually the parser expands $foo
to "foo" $, and @foo to "foo" @; $ and @ are special words).
- words are defined at eval time using a 'define' word, for example:
"squared" [ dup * ] define
- continuations (!)
I will release the interpreter very soon, as open source; the release of
the game itself is still some ways off.
> 3. How are sophisticated data structures best
> modelled in Joy?
I'm not sure its possible in Joy itself. However it is not hard to add
this concept to an RPN language; my language supports 'namespaces',
which are dictionaries used to look up word and variable definitions.
Namespaces can be nested, and used to model data structures -- for
example, each object in the game has an associated namespace with
variables x, y, z, dx, dy, dz, and so on. An object system is built on
top of this -- a new instance is created by cloning an existing
namespace, inheritance is done simply by cloning a new instance and
adding words and variables. The ; operator is used to evaluate a
quotation inside a namespace:
$player [ $x $y ] ;
> 4. Do people see concatenative languages as
> THE way forward, or do they see them as a good
> tool to use for some types of programming?
While some people hate RPN, I love it how most definitions fit on a few
lines, and those that don't are easy to break up into multiple
definitions.
--
Slava Pestov
phimvt@lurac.latrobe.edu.au — 2003-11-19 06:20:11
On Tue, 18 Nov 2003, Mark Phillips wrote:
[..]
> 1. Is Joy an exploratory language only? Could it
> be used in a real-world application? If not, how
> far away from this is it? Could it be used in
> combination with another language to make it
> useable?
Well, it started off as an exploration. By now it could be used for
real world problems by Joy programmers. But I cannot at this stage
see application programs written in it.
> 2. Is there potential eventually for a sophisticated
> type system as part of Joy? Would any real typing
> system have to be modelled within the Joy language
> itself? Is this a reasonable thing to do?
In this respect Joy is like Lisp and Scheme, it is weakly and
dynamically typed. I don't know what you mean by "real typing"
(statically ?) or by "modelled within ..". There is no reason
why somebody could not design and implement a statically typed
language similar to Joy, but it would be as different as ML and
Haskell are from Lisp and Scheme.
> 3. How are sophisticated data structures best
> modelled in Joy?
As in Lisp and Scheme: nested lists.
> 4. Do people see concatenative languages as
> THE way forward, or do they see them as a good
> tool to use for some types of programming?
Well, for functional programming the concatenative model
is one alternative. "THE way .." - who knows ?
> 5. Do notions of parallel programing fit within
> Joy's model?
Yes, all functional programming languages lend themselves to
parallel implementations. Very few have kept that promise,
and most, Joy included, have not even tried.
> 6. What, if any, are the downsides of Joy's
> approach?
It's new, it's very different, even when compared with other
functional languages. So the prime downside is conceptual inertia,
mostly. Also, many people would argue that functional programming
is too narrow.
Hope this helps.
- Manfred
John Cowan — 2003-11-19 14:01:25
Mark Phillips scripsit:
> 1. Is Joy an exploratory language only? Could it
> be used in a real-world application? If not, how
> far away from this is it? Could it be used in
> combination with another language to make it
> useable?
Joy 1 has access to almost the whole standard library of ISO C.
At present, the biggest limitation is that there's no way to
load foreign routines into the Joy interpreter.
> 2. Is there potential eventually for a sophisticated
> type system as part of Joy? Would any real typing
> system have to be modelled within the Joy language
> itself? Is this a reasonable thing to do?
Well, Joy doesn't have variables, so it isn't going to have
static typing. Dynamic typing already exists, although the
list of types (int, float, string, stream, list, set of integers, symbol)
is currently fixed. It wouldn't be that hard to add an extensible
type system.
> 3. How are sophisticated data structures best
> modelled in Joy?
As lists.
> 4. Do people see concatenative languages as
> THE way forward, or do they see them as a good
> tool to use for some types of programming?
Hopefully the latter.
> 5. Do notions of parallel programing fit within
> Joy's model?
There's no reason why they shouldn't, and in fact it
would be fairly easy to add multiprocessing.
> 6. What, if any, are the downsides of Joy's
> approach?
It's hard to keep track / Of what's on the stack.
--
John Cowan
http://www.ccil.org/~cowan <
jcowan@...>
You tollerday donsk? N. You tolkatiff scowegian? Nn.
You spigotty anglease? Nnn. You phonio saxo? Nnnn.
Clear all so! `Tis a Jute.... (Finnegans Wake 16.5)
Chris Double — 2003-11-19 18:43:18
On Wed, 19 Nov 2003 17:20:11 +1100,
phimvt@... said:
>
> In this respect Joy is like Lisp and Scheme, it is weakly and
> dynamically typed.
Just a nitpick, Scheme and Common Lisp are strongly typed, not weakly
typed. They are dynamically typed though. So they are classified as
having strong/dynamic typing. This is compared to C and C++'s weak/static
typing model.
Chris.
--
http://radio.weblogs.com/0102385
--
Chris Double
chris.double@...
Tanksley, William D. Jr. — 2003-11-19 19:02:22
From: John Cowan [mailto:
cowan@...]
>Mark Phillips scripsit:
>Well, Joy doesn't have variables, so it isn't going to have
>static typing.
Variables aren't needed for static typing -- strongForth has strong static
typing without variables. Joy itself won't have static typing, but a dialect
of it could. I've been designing one such for some time, although work on it
has been nonexistant.
>John Cowan http://www.ccil.org/~cowan
-Billy
Tanksley, William D. Jr. — 2003-11-19 19:11:24
From: Mark Phillips [mailto:
pierryth@...]
>2. Is there potential eventually for a sophisticated
>type system as part of Joy?
Yes. See strongForth (Google) for an example and explanation of how to do
this.
> Would any real typing
>system have to be modelled within the Joy language
>itself?
No. It would be best modelled as part of the compiler.
>Is this a reasonable thing to do?
I say yes. Others say no.
>3. How are sophisticated data structures best
>modelled in Joy?
As lists.
>4. Do people see concatenative languages as
>THE way forward, or do they see them as a good
>tool to use for some types of programming?
Good tool, I would say -- and also a great avenue for research, since
they've been so sparsely studied.
>5. Do notions of parallel programing fit within
>Joy's model?
Yes and no. Yes, because it's a functional language; no, because the
operations Joy provides have an innate sequencing. cK is a derivative of Joy
which could be made parallel much easier, since it includes and heavily uses
array operations, which often have no explicit ordering.
http://www.nsl.com/papers/ck.htm
My ideal language is a mixture of cK and strongForth.
>6. What, if any, are the downsides of Joy's
>approach?
Probably the worst downside is that we don't know the downsides or upsides.
This type of language is old and well-used (Forth and Postscript), but very
poorly researched. It's also a downside that few professional programmers
have experience with anything like it.
>Mark.
-Billy
Paul-V Khuong — 2003-11-19 23:37:46
> Message: 3
> Date: Wed, 19 Nov 2003 17:20:11 +1100
> From: phimvt@...
> Subject: Re: Questions about Joy
>
>
> On Tue, 18 Nov 2003, Mark Phillips wrote:
[...]
> > 2. Is there potential eventually for a
> sophisticated
> > type system as part of Joy? Would any real typing
> > system have to be modelled within the Joy language
> > itself? Is this a reasonable thing to do?
>
> In this respect Joy is like Lisp and Scheme, it is
> weakly and
> dynamically typed.
I've absolutely no experience in Joy (school is eating
my time up), but i'd like to exit lurk mode and say
that Lisp (and Scheme, afaik) are STRONGLY but
dynamically typed.
__________________________________
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
http://antispam.yahoo.com/whatsnewfree
stevan apter — 2003-11-20 01:55:18
i think the best way to learn about joy is just to
jump in and start coding.
some years ago i heard ken iverson (the inventor of
APL) urge people to pick a textbook which uses C or
Java and rewrite the examples in APL. i seem to
recall that he had plans to J-ify knuth's _concrete
mathematics_:
http://www.amazon.com/exec/obidos/tg/detail/-/0201558025/qid=1069292917/sr=8-1/ref=sr_8_1/103-7469654-4278259?v=glance&n=507846
this would make a terrific joy project.
----- Original Message -----
From: "Mark Phillips" <pierryth@...>
To: <concatenative@yahoogroups.com>
Sent: Tuesday, November 18, 2003 11:17 PM
Subject: [stack] Questions about Joy
> Hi,
>
> I've just heard about Joy and it sounds like a
> very interesting language. I have a few questions
> which I'm hoping people can shed some light on.
>
> 1. Is Joy an exploratory language only? Could it
> be used in a real-world application? If not, how
> far away from this is it? Could it be used in
> combination with another language to make it
> useable?
>
> 2. Is there potential eventually for a sophisticated
> type system as part of Joy? Would any real typing
> system have to be modelled within the Joy language
> itself? Is this a reasonable thing to do?
>
> 3. How are sophisticated data structures best
> modelled in Joy?
>
> 4. Do people see concatenative languages as
> THE way forward, or do they see them as a good
> tool to use for some types of programming?
>
> 5. Do notions of parallel programing fit within
> Joy's model?
>
> 6. What, if any, are the downsides of Joy's
> approach?
>
> Thanks,
>
> Mark.
>
>
> __________________________________
> Do you Yahoo!?
> Protect your identity with Yahoo! Mail AddressGuard
> http://antispam.yahoo.com/whatsnewfree
>
>
> To unsubscribe from this group, send an email to:
> concatenative-unsubscribe@egroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
Michael Manti — 2003-11-20 02:08:10
Concrete Math Companion is at
http://www.jsoftware.com/publications_books.htm
On Wednesday, November 19, 2003, at 08:55 PM, stevan apter wrote:
> i think the best way to learn about joy is just to
> jump in and start coding.
>
> some years ago i heard ken iverson (the inventor of
> APL) urge people to pick a textbook which uses C or
> Java and rewrite the examples in APL. i seem to
> recall that he had plans to J-ify knuth's _concrete
> mathematics_:
>
> http://www.amazon.com/exec/obidos/tg/detail/-/0201558025/
> qid=1069292917/sr=8-1/ref=sr_8_1/103-7469654-4278259?v=glance&n=507846
>
> this would make a terrific joy project.
>
>
> ----- Original Message -----
> From: "Mark Phillips" <pierryth@...>
> To: <concatenative@yahoogroups.com>
> Sent: Tuesday, November 18, 2003 11:17 PM
> Subject: [stack] Questions about Joy
>
>
> > Hi,
> >
> > I've just heard about Joy and it sounds like a
> > very interesting language. I have a few questions
> > which I'm hoping people can shed some light on.
> >
> > 1. Is Joy an exploratory language only? Could it
> > be used in a real-world application? If not, how
> > far away from this is it? Could it be used in
> > combination with another language to make it
> > useable?
> >
> > 2. Is there potential eventually for a sophisticated
> > type system as part of Joy? Would any real typing
> > system have to be modelled within the Joy language
> > itself? Is this a reasonable thing to do?
> >
> > 3. How are sophisticated data structures best
> > modelled in Joy?
> >
> > 4. Do people see concatenative languages as
> > THE way forward, or do they see them as a good
> > tool to use for some types of programming?
> >
> > 5. Do notions of parallel programing fit within
> > Joy's model?
> >
> > 6. What, if any, are the downsides of Joy's
> > approach?
> >
> > Thanks,
> >
> > Mark.
> >
> >
> > __________________________________
> > Do you Yahoo!?
> > Protect your identity with Yahoo! Mail AddressGuard
> > http://antispam.yahoo.com/whatsnewfree
> >
> >
> > To unsubscribe from this group, send an email to:
> > concatenative-unsubscribe@egroups.com
> >
> >
> >
> > Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/
> >
> >
>
>
<image.tiff>
>
>
> To unsubscribe from this group, send an email to:
> concatenative-unsubscribe@egroups.com
>
>
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
Michael Manti
mmanti@...
[Non-text portions of this message have been removed]
stevan apter — 2003-11-22 16:50:54
----- Original Message -----
From: "Mark Phillips" <pierryth@...>
To: <concatenative@yahoogroups.com>
Sent: Tuesday, November 18, 2003 11:17 PM
Subject: [stack] Questions about Joy
> Hi,
>
> I've just heard about Joy and it sounds like a
> very interesting language. I have a few questions
> which I'm hoping people can shed some light on.
>
> 1. Is Joy an exploratory language only? Could it
> be used in a real-world application?
you might want to cut your joy teeth on this problem:
http://www.ipd.uka.de/~prechelt/phonecode/
over the last few days there's been some discussion of
this problem on comp.lang.forth, but - alas - no code.
i've got one in k (http://www.nsl.com/papers/phone.htm),
and that algorithm is easily translated into cK, but a
pure joy solution would succeed in getting the attention
of the forth community. i'm particularly struck by the
naturalness of the recursive solution, so on that score
it would seem that joy has something uniquely interesting
to contribute.
any takers?
S. Alexander Jacobson — 2003-12-03 03:12:00
On Wed, 19 Nov 2003
phimvt@... wrote:
> > 1. Is Joy an exploratory language only? Could it
> > be used in a real-world application? If not, how
> > far away from this is it? Could it be used in
> > combination with another language to make it
> > useable?
>
> Well, it started off as an exploration. By now it could be used for
> real world problems by Joy programmers. But I cannot at this stage
> see application programs written in it.
Ok, I'll bite.
How would you write an httpd service in Joy?
Does it have a concurrency model?
How does one deal w/ a slow connection?
> > 5. Do notions of parallel programing fit within
> > Joy's model?
>
> Yes, all functional programming languages lend themselves to
> parallel implementations. Very few have kept that promise,
> and most, Joy included, have not even tried.
Ok. Can you sketch this?
-Alex-
___________________________________________________________________
S. Alexander Jacobson Check out my new blog!!!
1-212-787-1914 voice
http://alexjacobson.com
Nick Forde — 2003-12-03 10:36:11
Hi Alex,
S. Alexander Jacobson writes:
> On Wed, 19 Nov 2003 phimvt@... wrote:
> > > 1. Is Joy an exploratory language only? Could it
> > > be used in a real-world application? If not, how
> > > far away from this is it? Could it be used in
> > > combination with another language to make it
> > > useable?
> >
> > Well, it started off as an exploration. By now it could be used for
> > real world problems by Joy programmers. But I cannot at this stage
> > see application programs written in it.
>
> Ok, I'll bite.
> How would you write an httpd service in Joy?
> Does it have a concurrency model?
> How does one deal w/ a slow connection?
Currently to do the above it would be necessary to extend the
interpreter, ideally through a general mechanism to support foreign
language integration.
Personally, I wouldn't recommend using Joy for any sizeable real-world
application until there is an industrial strength interpreter
available and associated libraries. However, having said this I have
used it for scripting various admin. jobs and this has let me use it
in a professional environment.
You will find Joy rough around the edges compared to languages such as
Ruby, Python and Perl, but hey, its fun writing Joy code and worth
giving a go!
Nick.
phimvt@lurac.latrobe.edu.au — 2003-12-08 03:00:25
On Tue, 2 Dec 2003, S. Alexander Jacobson wrote:
> On Wed, 19 Nov 2003 phimvt@... wrote:
> > > 5. Do notions of parallel programing fit within
> > > Joy's model?
> >
> > Yes, all functional programming languages lend themselves to
> > parallel implementations. Very few have kept that promise,
> > and most, Joy included, have not even tried.
>
> Ok. Can you sketch this?
All functional languages are based on the evaluation of
expressions. Often expressions contain subexpressions which
can be evaluated in any order, and even in parallel, without
any effect on the final result. An example, written in
infix and in Joy notation:
a) (1 + 2) * (7 - 3) 1 2 + 7 3 - *
b) 3 * 4 3 4 *
c) 12 12
Here a) is the original expression, in step b) the addition
and the subtraction are performed "in parallel", and in step
c) the final multiplication is performed. This is similar to
the way we wrote expression evaluation at school, except that
we really did the addition first and then the subtraction.
But in a truly parallel implementation the two operations
in b) would be done at the same time, by separate brains or
separate processors. I do not know of any implementations
that actually do use separate processors, but I believe some
of the optimisations that are done on the fly by some of
the latest hardware come pretty close. Not my field, though.
Hope this helps.
- Manfred